Skip to content

is a machine learning project for forecasting steel prices using historical data. It employs Prophet for time-series analysis, supports additional regressors, and includes robust data preprocessing

Notifications You must be signed in to change notification settings

TouHouE/SteelPraicePredict

Repository files navigation

How to install package:

由於Prophet套件的安裝方式在windows與Linux上略有不同,因此詳細安裝方式請參考官方github

How to use:

predict.py

    python predict.py --start_date "2020-09-20" --predict_range "10D" --predicted_range "./for_iamge"

What meaning with those arguments:

    --start_date:   預測起始日期,請勿超出DataBase裡的最後日期。
                 格式為"年年年年-月月-日日"。必須要加字串引號

    --predict_range:    預測未來時間的長度,後半部分為時間單位為天(D)、週(W)或月(M)
                    前半部分表示長度。必須要加字串引號
    --predicted_dir:   用來儲存預測數據及歷史數據的資料夾,若是要Demo趨勢圖請指定位置。必須要加字串引號

showDemo.py

     python showDemo.py --predict_csv ./for_image/djusst_D.csv --history_csv ./for_image/djusst_D_DF.csv --image_name djusst_D.jpg

What meaning with those arguments:

     --predict_csv:    預測數據的.csv表位置及檔名。無需使用字串引號
     --history_csv:    歷史數據的.csv表位置及檔名。無需使用字串引號
     --image_name:     使用這個名字儲存圖檔,若是不加則是執行完後跳出可互動視窗的顯示方式。無需使用字串引號

Requirement

  1. sciki-learn
    • version: 0.23.2
  2. pystan
    • version: 2.19.1.1
  3. fbprophet
    • version: 0.7.1
  4. pandas
    • version: 1.3.1
  5. joblib
    • version: 1.0.1
  6. pymysql
    • version: 1.0.2

TODO: 修正期始日期不可超過DataBase中最後的日期

About

is a machine learning project for forecasting steel prices using historical data. It employs Prophet for time-series analysis, supports additional regressors, and includes robust data preprocessing

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages